NASA Exploring Generative AI Tool With Industry Partners

NASA and its industry partners are testing a generative artificial intelligence capability that could have access to the agency’s sensitive internal data in the future.

Jeff Seaton, chief information officer at NASA, said the agency expects to have the large language model rated at a Federal Information Security Modernization Act Moderate level by mid- to late summer to enable using such data.

The LLM test is part of NASA’s effort to adopt more advanced AI tools that can support the technical side of missions. At present, the agency uses AI-enabled capabilities for tracking wildfire smoke, counting trees and identifying the location of exoplanets, Nextgov/FCW reported.

The test will ensure that the generative AI tool will be used in a safe and responsible manner in accordance with an executive order on AI.

According to David Salvagnini, NASA’s chief data and AI officer, the agency understands that advanced technologies could produce biased results due to training data and is working to ensure the issue is addressed before integrating new tools into operations.

“We can make our work more efficient, but that’s only if we approach these new tools in the right way, with the same pillars that have defined us since the beginning: safety, transparency and reliability,” said NASA Administrator Bill Nelson, a 2024 Wash100 awardee.